Hey there guys !
I was just starting to think about implementing OZ Defender for a well-known project, but I'm realizing that maybe using defender via UI its cool - but it does not scale well for some stuff.
There are two cases that come into mind:
- I want to setup a Sentinel that alerts me when a job in Keep3r network is being griefed. For this, I'd have to setup a condition that checks on the calls of my job contract, and I'd want to configure something like: 'if this function has been called with parameter P an X amount of times within Y interval do Z'.
The thing is ... I want P to be diff. parameters, so in order to achieve this, I have to do this manually, for like 50 parameters.
- I want to be able to setup all of this, and have some version control / share with my peers / reuse setups / etc. for this, ofc, github is a must.
My take
I think that if OZ Defender had some kind of CLI/API to create / modify sentinels (and then scale to autotasks, etc.) with an easy-to-understand configuration format (YML/JSON/etc.). It would allow a lot of cool stuff and implementations for big projects, also, it would allow CI integrations (deployment, testing, etc.)
Also, if it would be manageable to create a local OZ defender ecosystem (with a docker image?), one could also do some integration testing of sentinels and autotasks with mainnet forks.
It would be the holy grail of chaotic / stress testing for triage teams.